Large Language Model Results: NVIDIA
The GPT3 results are the first time we have seen apples-to-apples benchmarks for training the large language models that have recently taken AI from an esoteric tech to superstar status. Since training the full GPT3 costs about $10M and takes months to run, the MLCommons community came up with a subset that is both indicative of how well a platform performs on the full 185B parameter model, and that could be run at a reasonable cost and wall time.
NVIDIA is really good at running these benchmarks, and as has always been the case, they ran every benchmark. For the GPT3 benchmark, NVIDIA only submitted their latest and greatest GPU, the H100, depending on their partners to run on the A100.
Maybe this is one reason why AMD has been drifting down since it hit $128 - no product to test…doc
5 Likes
This jumped out regarding price volatility: As such, TSMC’s focus on ramping up its advanced packaging capacity for its design partners is critical, as there were suggestions that TSMC could lack sufficient ability to meet near-term AI chip demand needs. As such, it’s possible that the elevated expectations over AMD and Nvidia’s growth inflection could have been overstated, leading to a steeper pullback if the market anticipates the earnings accretion might not pan out in the second half.
AMD has been good at executing.
While AMD reportedly launched superior AI chip hardware specifications than Nvidia’s H100, whether its open-source software ecosystem could compete effectively against Nvidia’s proprietary approach remains to be seen. For now, I assessed that these uncertainties would likely dampen further upward re-rating on AMD’s valuation. However, that also suggests that it could provide an attractive opportunity for high-conviction investors who have faith in AMD’s execution in working out these challenges and gaining share against Nvidia.
These guys raised AMD to a buy, but in the disclaimer they also hold a position in AMD…doc
1 Like
Forbes has 2 article in the last day on AMD, both somewhat negative.
Now, should you buy or fear AMD stock at the current market price of about $111 per share? While there are a couple of tailwinds for the stock, we think AMD’s valuation at the current juncture appears a bit rich. AMD stock trades at about 8x sales, well ahead of the broader market. Moreover, although the company has seen robust growth (52% on average, over the last three years) driven by strong PC demand through the Covid-19 pandemic and a growing uptake of data center chips by cloud players, revenues are poised to contract this year as demand cools off. AMD’s operating margins have also contracted considerably to -3% over the most recent quarter, compared to about 16% in the year-ago period. AMD stock is also somewhat less resilient to downturns, as the stock has seen bigger drawdowns compared to the broader market over both the 2008 recession as well as the 2020 Covid-19 recession. We estimate AMD valuation at about $84 per share, which is 24% below the current market price.
Next maybe this is why they have sent any samples to be tested from the first article above…doc
The company expects to begin offering the technology for customers to try in Q3 2023, with production slated to ramp up from 2024. The company estimates that the total addressable market for data center AI accelerators will approach $150 billion by 2027, from roughly $30 billion this year.
4 Likes
NVDA current top favorite of the momo crowd… they’d better not disappoint. AMD has been doing well… both eating INTC’s lunch. Plenty of room for both of them long term.
2 Likes